ALPHA RELEASE / v0.1.0
Turn documentation into linked Markdown.
knowledge reads a URL or local HTML/Markdown file, splits it by headings, and asks a selected language model to extract one concept per section. The result is a directory of Markdown files you can read, review, and version in Git.
Python 3.12+ MIT license Configure an LLM provider
THE BASICS
A small pipeline with readable output.
- 01
Choose a source
Pass a URL, or a local HTML or Markdown file. URL fetching has a 50 MiB body limit.
- 02
Extract by heading
HTML headings from h2 to h4 and Markdown h2 sections become separate extraction requests.
- 03
Review the bundle
The selected model returns structured concepts. The SDK writes Markdown with YAML metadata and validates index links.
GETTING STARTED
Install, configure, run.
Install the package, set credentials for your chosen hosted provider, then point the CLI at a document. The default model is gpt-4o and requires OPENAI_API_KEY; the guide covers other providers and the Python API.
$ pip install knowledge
$ knowledge create https://example.com/docs.html ./docs-bundle
PROJECT STATUS
This is an alpha release.
The package is classified as Alpha. APIs and the bundle format may change; review the changelog and open issues before depending on it in a production workflow.